The german site "SAS und Chiffrierdienst der DDR" http://scz.bplaced.net has some interesting content under point 11.3 in the category "manuelle Verfahren" I found ciphers named from A to D. But I think none of these are secure besides security trough obscurity or am I wrong?
When looking at it it looked quite similar to mondern cryptographíc algorithms relying on a secret key and do some "magic" on the plain text to get the ciphertext.
So, on first thought I'd say that these codes follow the rule that it solely relies on the secrecy of a key (in the example on the page you linked it's
SCHULPFORTA DEGEN) and not the secrecy of the algorithm (which is the contrary of "security by obscurity").
I can't give analysises about the strongness of an algorithm by myself but because David Kahn's "Codebreakers" is mentioned and I have this book sitting in my shelf next to my computer, I can cite from there:
Such an encipherment virtually obliterates ciphertext repetitions. Even if an exact 18-letter plaintext group recurs, it must begin at exactly the same point in the encipherment equation to produce a ciphertext repetition - and there is only one chance in 18 of this happening. More importantly, a polygraphic encipherment of this magnitude is possible only with a Hill transformation. The more than 40 quintillion 18-letter groups, printed 100 to each side of a page with their ciphertext equivalents, would fill a codebook thicker than the distance from the sun to Pluto.
The cryptology of today is saturated with mathematical operations, mathematical methods, mathematical thinking. In practice, it has become virtually a branch of applied mathematics. Its sophistication, its range, and its power have grown far beyond the imaginings of the most imaginative cryptologist in Yardley's Black Chamber. And in this evolution, Lester Hill was a prime mover.
BTW: By "today" Kahn is speaking about the 60s/70s, the book was publshed 1967.
Edit: As promised I had a look at C1 and C2.
At first glance at C1 I thoght the same but there is a bit more to it than "security by obscurity". C1 uses a key and one of five filling methods. The key not only sets up the substitution table but is also used to specify the order in which to write down the blocks of cipher elements after filling the key-table.
So an attacker who wants to do statistical analysis to break the key needs to guess the used filling method, the length of the key and the order of letters in which they appear in the key. A brute force attack today might not take that long with todays computing power but for the 50s that's another story.
C2 on the other hand doesn't look robust even after second glance. There are only four substution tables (all of which leaves a cipher that is vulnerable to statistic attacks) and a fixed filling method, so by knowing about this algorithm, breaking a cipher shouldn't be hard.
But even C2 might have it's use because in the Cold War (and any other time that is ;-) you sometime want your enemy to "break" a code in order to feed false information. The information in a deciphered message looks more authentic than a plain message you find on a postcard.
Also, it might also be suitable for messages where the value of the information decreases very fast. E.g. if I use C2 as cipher to send a buying order to my broker and it keeps other trading parties from reading that information until the order is fulfilled (and the information can then be found in the order book anyway) this cipher (which can be calculated manually quite fast) might still be suitable.